See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/12202552

Personality and Job Performance: The Big Five Revisited

Article in Journal of Applied Psychology · December 2000

DOI: 10.1037/0021-9010.85.6.869 · Source: PubMed

CITATIONS READS

1,642 50,710

2 authors:

Gregory M Hurtz John Donovan

California State University, Sacramento Rider University
15 PUBLICATIONS 2,317 CITATIONS 13 PUBLICATIONS 3,159 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by John Donovan on 02 September 2014.

The user has requested enhancement of the downloaded file.

Journal of Applied Psychology Copyright 2000 by the American Psychological Association, Inc.
2000, Vol. 85, No. 6, 869-879 0021-9010/00/$5.00 DOI: 10.1037//0021-9010.85.6.869

Personality and Job Performance: The Big Five Revisited

Gregory M. Hurtz John J. Donovan

University at Albany, State University of New York Virginia Polytechnic Institute and State University

Prior meta-analyses investigating the relation between the Big 5 personality dimensions and job
performance have all contained a threat to construct validity, in that much of the data included within
these analyses was not derived from actual Big 5 measures. In addition, these reviews did not address the
relations between the Big 5 and contextual performance. Therefore, the present study sought to provide
a meta-analytic estimate of the criterion-related validity of explicit Big 5 measures for predicting job
performance and contextual performance. The results for job performance closely paralleled 2 of the
previous meta-analyses, whereas analyses with contextual performance showed more complex relations
among the Big 5 and performance. A more critical interpretation of the Big 5-performance relationship
is presented, and suggestions for future research aimed at enhancing the validity of personality predictors
are provided.

During the several decades prior to the 1990s, the use of 1990), researchers in the early 1990s began to adopt this Big Five
personality testing in employee selection was generally looked framework for selection research (Barrick & Mount, 1991; Tett,
down on by personnel selection specialists. This was primarily due Jackson, & Rothstein, 1991).
to pessimistic conclusions drawn by researchers such as Guion and Early meta-analytic work by B arrick and Mount (1991) and Tett
Gottier (1965) in their qualitative review of the personality testing et al. (1991) provided evidence suggesting that the Big Five might
literature and by Schmitt, Gooding, Noe, and Kirsch (1984) in their have some degree of utility for selecting employees into a variety
quantitative meta-analysis of various personnel selection tech- of jobs. In both of these reviews, the researchers used studies that
niques. The general conclusion drawn by these researchers was provided correlations between any type of personality variable and
that personality tests did not demonstrate adequate predictive job performance, categorizing the various personality variables
validity to qualify their use in personnel selection. In fact, Schmitt into one of the Big Five dimensions to estimate the strength of
et al. (1984) found that personality tests were among the least valid these variables' correlation with job performance. Although their
types of selection tests, with an overall mean sample-size weighted results were not altogether consistent (see Ones et al., 1994, and
correlation of .21 for predicting job performance, and concluded Tett, Jackson, Rothstein, & Reddon, 1994, for a discussion of
that "personality tests have low validity" (p. 420). reasons), the general consensus drawn by researchers and practi-
Over the past several years, however, there has been an in- tioners was that personality does in fact hold some utility as a
creased sense of optimism regarding the utility of personality tests predictor of job performance. The impact of these studies on
in personnel selection (Behling, 1998; Goldberg, 1993; Hogan, raising the status of personality tests in employee selection has
Hogan, & Roberts, 1996; Hogan & Ones, 1997; Mount & Barrick, been felt throughout the 1990s. Subsequent meta-analyses by
1995). In recent years, researchers have suggested that the true Mount and Barrick (1995) and Salgado (1997) have seemed to
predictive validity of personality was obscured in earlier research solidify this newfound status granted to personality, particularly to
by the lack of a common personality framework for organizing the Conscientiousness. Behling (1998), for example, recently claimed
traits being used as predictors (Barrick & Mount, 1991; Hough, Conscientiousness as one of the most valid predictors of perfor-
1992; Mount & Barrick, 1995; Ones, Mount, Barrick, & Hunter, mance for most jobs, second only to general intelligence.
1994). With increasing confidence in the robustness of the five- Much of the recent enthusiasm for the Big Five in personnel
factor model of personality (Digman, 1990; Goldberg, 1993; John, selection has been based on this body of meta-analytic work,
especially the original work of Barrick and Mount (1991). In fact,
on the basis of this work, most researchers seem satisfied to
Gregory M. Hurtz, Department of Psychology, University at Albany, conclude that Conscientiousness is a generally valid predictor of
State University of New York; John J. Donovan, Department of Psychol- job performance and that it represents the primary, if not the sole,
ogy, Virginia Polytechnic Institute and State University. personality dimension for use in personnel selection. We feel that
A version of this study was presented at the 13th Annual Conference of it is necessary to revisit and explore the Big Five in this domain for
the Society for Industrial and Organizational Psychology, Dallas, Texas,
three main reasons.
April 1998. We thank Kevin Williams, Stephen Dwight, and Jesfs Salgado
First, we feel that there are methodological and statistical issues
for their helpful comments and suggestions.
Correspondence concerning this article should be addressed to Gregory pertaining to past meta-analytic reviews that warrant a critical
M. Hurtz, Department of Psychology, Social Sciences 112, University at reanalysis of the research literature that is commonly cited as
Albany, State University of New York, Albany, New York 12222. Elec- supporting the criterion-related validity of the Big Five. Second, as
tronic mail may be sent to [email protected]. several years have passed since the Big Five was adopted as the

869
870 HURTZ AND DONOVAN

dominant personality framework for personnel selection, we feel it sions. When faced with multiple scales categorized into the same
would be beneficial to meta-analyze this body of research in which dimension from a single study, Barrick and Mount (1991) entered
actual measures of the Big Five were correlated with job perfor- the average correlation across these scales into their meta-analysis.
mance. Third, given recent developments in the research explicat- Tett et al. (1991) entered the average absolute value correlation in
ing the job performance criterion domain (e.g., Borman & Moto- such instances. As Mount and Barrick (1995) noted, using the
widlo, 1993, 1997; Motowidlo & Van Scotter, 1994; Van Scotter average correlation underestimates the validity of the higher order
& Motowidlo, 1996), we feel it would be beneficial to meta- construct to which these scales purportedly belong. Instead, a
analytically explore the relations between the Big Five and these composite score correlation should be computed to reflect the
various dimensions of job performance. correlation between the sum of the lower order constructs and the
criterion. Mount and Barrick (1995) and Salgado (1997) used this
M e t h o d o l o g i c a l and Statistical Issues in Past R e v i e w s composite-score correlation procedure and demonstrated a result-
ing increase in the estimated validities of the Big Five. However,
With respect to prior meta-analytic work examining the utility the fact still remains that these are only estimates of the validities
of the Big Five in personnel selection, we feel that there are two of actual Big Five measures, because these researchers' studies did
main weaknesses in these reviews that need to be addressed prior not exclusively include correlations from actual Big Five mea-
to making conclusions about the use of personality for personnel sures. Thus, the degree to which these meta-analyses have pro-
selection. First, it appears that all four major meta-analyses pub- vided accurate estimates of the "true" validities of the actual Big
lished up to this point (Barrick & Mount, 1991; Mount & Barrick, Five remains to be seen.
1995; Salgado, 1997; Tett et al., 1991) contain a potential threat to If we accept these previous estimates of the relation between the
construct validity resulting from the methods the researchers used Big Five and job performance, our second concern then centers
to derive their meta-analytic estimates of criterion-related validity. around the overwhelmingly positive interpretation of these esti-
This threat stems from the fact that these validity coefficients were mates. As we mentioned previously, Schmitt et al. (1984) sug-
largely based on studies that used measures that were not designed gested that a mean sample-size weighted observed correlation of
to explicitly measure the Big Five personality dimensions. Instead, .21 for personality, averaged across various personality scales
all four of these reviews were based on data from a diverse without a unified framework, indicated that personality has low
collection of non-Big Five measures that were classified post hoc validity for predicting job performance. Consistent with this con-
into the Big Five categories. Although these were gallant efforts at clusion, the selection community generally looked down on the use
addressing the relation between the Big Five and job performance of personality as a means of predicting job performance. We find
given the limited data available in the literature at that time, this it curious that a number of years later, after the Big Five frame-
post hoc classification procedure has raised some concern in the work was adopted in subsequent recta-analyses, there have been
personnel selection research community over the validity of the such positive conclusions concerning the criterion-related validity
results obtained in these past reviews (Hogan et al., 1996; Ones et of Conscientiousness, given that the mean sample-size weighted
al., 1994; Salgado, 1997; Tett et al., 1994). observed correlations for Conscientiousness were lower than that
The central issues concerning this classification procedure are found by Schmitt et al. (1984; the mean sample-size weighted
the suboptimal levels of interrater agreement in the classification observed ?s for Conscientiousness ranged from. 10, Salgado, 1997,
of the various personality scales into the Big Five dimensions and to .18, Mount & Barrick, 1995, in these later meta-analyses). In
the misclassification of some scales into these dimensions. An fact, even Barrick and Mount's (1991) estimate of the true corre-
inspection of the methods reported by both Barrick and Mount lation for Conscientiousness, after corrections for range restriction
(1991) and Tett et al. (1991) reveals that the level of interrater and unreliability in both the predictors and criteria, was approxi-
agreement achieved within each of these reviews is not entirely mately equal to Schmitt et al.'s uncorrected estimate. Despite these
satisfactory. For example, Barrick and Mount ( 1991) reported only facts, these later reviews met with immediate enthusiasm for the
83% or better rater agreement on 68% of the classifications, potentially valuable role of Conscientiousness in selection.
suggesting less than desirable interrater agreement. In light of such In our view, this enthusiasm has resulted from two forces. First,
difficulties in agreeing on scale classifications, it is not entirely from a theoretical perspective, the Conscientiousness construct
unlikely that errors may have been made in these classifications. does seem to be logically related to j o b performance. It makes
As evidence of this problem, Hogan et al. (1996) found that a intuitive sense that individuals who have characteristic tendencies
number of errors had been made in how scales were classified in to be dependable, careful, thorough, and hardworking should be
these early meta-analyses. Additionally, Salgado (1997) indicated better performers on the job. It is therefore understandable that so
that the same scales bad been classified into different categories by much interest has arisen in this construct as it relates to employee
the different groups of researchers when they conducted their selection. Schmitt et al. (1984), on the other hand, had no specific
separate meta-analyses. He suggested that this situation arose construct to point to in their analysis, as their validity coefficient
because there is a degree of ambiguity about how several scales was obtained by combining results across a variety of personality
map onto the Big Five, making it difficult to assign them exclu- variables with no attempt at categorization.
sively to one dimension (Salgado, 1997). These facts raise some Second, we believe that these validity coefficients for the Big
questions about the accuracy of the classifications and about the Five have often been interpreted in relative rather than in absolute
degree to which the meta-analytic findings map onto the actual Big terms. That is, in these meta-analyses (with the exception of Tett
Five constructs. et al., 1991), Conscientiousness has emerged as the most valid of
An issue that is related to the classification of scales is the the Big Five, and this has often been interpreted as indicating that
methods used for aggregating validity coefficients within dimen- Conscientiousness is valid in an absolute sense. On the contrary,
 BIG FIVE AND JOB PERFORMANCE 871

three of the meta-analyses present estimated true correlations for yses showing the relations between the other Big Five dimensions
Conscientiousness ranging from .15 to .22 (including statistical and those various criteria. Finally, Tett et al. (1991) performed no
corrections for range restriction, predictor unreliability, and crite- moderator analyses for criterion types but instead included only
rion unreliability)--correlations that do not fare extremely well correlations computed between personality scales and the criterion
when compared to absolute standards that have been used in dimensions they were hypothesized to predict, and their results
related research. A meta-analysis by Iaffaldano and Muchinsky were more positive in terms of the impact of Big Five factors other
(1985), for example, obtained a correlation of .17 between job than Conscientiousness on job performance.
satisfaction and job performance; this finding has been widely The findings of Tett et al. (1991) and Mount and Barrick (1995)
cited as indicating that there is no meaningful relationship between do provide some evidence that the link between the Big Five and
these constructs. Similarly, Cohen (1988) suggested .20 as an job performance might be more complex than has recently been
approximate standard that should be met for relationships between suggested, in that their degrees of validity depend on careful
constructs to be considered meaningful. Furthermore, as we noted selection of theoretically relevant criterion dimensions. Recent
previously, Schmitt et al. (1984) concluded that a correlation of .21 work by Motowidlo and Van Scotter (1994; Van Scotter & Mo-
was too low to consider personality a useful predictor of job towidlo, 1996) has likewise indicated that the Big Five have
performance. Finally, Mount and Barrick (1995) raised the stan- differing relations with theoretically linked dimensions of job
dards even further by suggesting that validities below .30 are performance within the task-versus-contextual distinction expli-
questionable, given the wide range of more valid predictors we cated by Borman and Motowidlo (1993, 1997). This body of work
have to choose from. has suggested that personality predictors should have their largest
If we were to adopt this .30 standard, only Mount and Barrick impact on contextual dimensions of job performance. Van Scotter
(1995) have provided evidence that Conscientiousness may be a and Motowidlo (1996) showed further that Extraversion and
valid predictor of job performance in an absolute sense. Whereas Agreeableness were more strongly related to the interpersonal
Barrick and Mount (1991) and Salgado (1997) found the estimated facilitation component of contextual performance than they were
true correlations between Conscientiousness and job performance to task performance. Although the magnitudes of these correlations
to be .22 and .25, respectively, Mount and Barrick (1995) found an were rather small, this finding does suggest that perhaps the Big
estimated overall true validity of .31. It is likely that this higher Five dimensions other than Conscientiousness take on importance
true validity is due to Mount and Barrick's use of composites score for predicting certain dimensions of job performance--a finding
correlations, as we discussed previously (Mount & Barrick, 1995). that may have been masked in the earlier meta-analyses. Thus, we
However, Salgado's lower estimate of .25 was based on the use of feel that the body of meta-analytic evidence relating the Big Five
composite score correlations as well and was also based on cor- to job performance would benefit from an exploration of their
rections for predictor unreliability that Mount and Barrick did not differential relations with task performance and the dimensions of
perform. Thus, in our view, these findings still do not give defin- contextual performance.
itive estimates of the true validities of explicit Big Five measures
and do not allow for confident conclusions regarding the validity S u m m a r y and Purpose
of Conscientiousness in an absolute rather than a relative sense. At
best, they indicate a low to moderate criterion-related validity for In summary, we are suggesting that the current body of meta-
Conscientiousness, despite recent enthusiasm that seems to suggest analytic work investigating the Big Five as predictors of job
a much stronger role for Conscientiousness in personnel selection performance contains some deficiencies that can now be ad-
(e.g., Behling, 1998). dressed. One major deficiency, in our view, is that all four of the
previous meta-analyses (i.e., Barrick & Mount, 1991; Mount &
D e v e l o p m e n t s in the Explication of the Job Performance Barrick, 1995; Salgado, 1997; Tett et al., 1991) suffer a potential
threat to construct validity in terms of the degree to which their
Criterion D o m a i n
predictors map onto the actual Big Five personality dimensions.
Another potential area in which the current body of meta- This methodological deficiency may have led to inaccurate esti-
analytic work can be improved on is the treatment of the criterion mates of the true relation between the Big Five and job perfor-
domain. Barrick and Mount (1991) performed a number of mod- mance. The current body of meta-analytic work in this area has
erator analyses for different types of criterion measures, and the provided general hypotheses about the strength of relation between
most clear finding was that their indicators of Conscientiousness the actual Big Five dimensions and job performance, suggesting
had a somewhat greater impact on subjective ratings than on that actual Big Five measures of Conscientiousness can be ex-
various types of objective ratings. The results for the other Big pected to produce criterion-related validities that are low to mod-
Five dimensions were less clear. Salgado (1997) split the criterion erate in magnitude.
domain into subjective ratings, personnel data, and training criteria In addition to overcoming this deficiency, we believe an explo-
and again found Conscientiousness to have a somewhat higher ration of the criterion-related validity of the Big Five for task
impact on subjective ratings than on objective criteria. Mount and versus contextual dimensions of job performance would aid in
Barrick (1995) were more careful to separate out dimensions of furthering this area of research. Motowidlo and Van Scotter (1994;
performance criteria that were theoretically meaningful with re- Van Scotter & Motowidlo, 1996) have begun to present evidence
spect to their relation with Conscientiousness, and they did find a in support of Big Five factors having differential validity with
pattern of differences showing Conscientiousness to relate to "will these different components of job performance. Thus, the purpose
do" or motivational factors more strongly than to "can do" or of the current study is both to meta-analytically summarize the
ability factors. However, Mount and Barrick did not present anal- body of research that has developed in recent years where actual
872 HURTZ AND DONOVAN

m e a s u r e s o f the Big Five were u s e d as predictors o f j o b perfor- classifiable into one of these categories because of mixed samples or
m a n c e and to test the criterion-related validities o f the Big Five for inadequate information. These studies were therefore excluded from this
theoretically relevant d i m e n s i o n s o f j o b performance. set of moderator analyses.
Criterion type. The type of criterion measure used when examining the
predictive validity of the Big Five was also coded as a potential moderator
Method of the personality-job performance relationship. The criterion domain was
analyzed in two separate ways. First. a two-category classification scheme
Literature Search was used, with the various criteria categorized as either measures of job
We used four separate methods to obtain validity coefficients for the proficiency or measures of training proficiency. Approximately 93% (42
present review. First, we conducted a computer-based literature search in out of 45) of the correlations were based on job proficiency criteria, and 37
PsycLit ( 1974 - 1996) and ERIC (1966 - 1996) using the key words person- of these were based on subjective ratings of job performance. Previous
ality and job performance, personality and training performance, five meta-analyses have analyzed subjective and objective performance mea-
factor model, and the Big Five. Second, we conducted a manual search in sures separately; in our data set, the objective analysis would have con-
the following journals tor the period of time from 1985 to 1998: Academy sisted entirely of objective sales data, making it a subset of studies from the
of Management Journal, Human Performance, Journal of Applied Psy- moderator analysis of the sales occupation. We theretbre decided to ex-
chology, Journal of Management, Journal of Personality and Social Psy- clude a separate moderator analysis for objective data. The training profi-
chology, Organizational Behavior and Human Decision Processes, and ciency category included both ratings of training performance and end-of-
Personnel Psychology. Third, we hand searched conference programs from training tests designed to evaluate learning and hands-on demonstration of
the last four annual conferences (1994-1997) of both the Society for skills. As very few training studies were found that used explicit Big Five
Industrial and Organizational Psychology (SIOP) and the Academy of measures, only 7% (3 out of 45) of the studies were based on measures of
Management for potential articles to be included in the present review. training proficiency.
Finally, we conducted a citation search in which the reference sections Second, we performed a separate analysis by partitioning the criterion
from previously gathered articles were examined to identify any potential domain into task performance, job dedication, and interpersonal facilitation
articles that may have been missed by earlier search methods. Using the (Motowidlo & Van Scotter, 1994; Van Scotter & Motowidlo, 1996). Using
selection criteria outlined below, we found 26 studies, yielding 35-45 definitions provided by Motowidlo and Van Scotter (1994) and Van
independent correlations for each of the Big Five dimensions. Scotter and Motowidlo (1996), we classified performance criteria such as
technical performance, use of equipment, job knowledge, completion of
specified job duties, and objective performance data as indicators of the
Criteria ,for Inclusion task performance category; ratings of work dedication, effort, persistence,
reliability, self-direction, commitment to objectives, and the like as indi-
For a study to be included in the present quantitative review, three
cators of the job dedication category; and ratings of interpersonal relations,
criteria had to be met. First, only studies using actual workers as partici-
cooperation, quality of interactions with others, being courteous, and being
pants in the research were included. Second, the study had to include a
a team player as indicators of interpersonal facilitation. Using these defi-
personality inventory that was explicitly designed from its inception to
nitions, we located within our sample 7-12 validity coefficients (across Big
measure the Big Five (i.e., the measure was constructed with the Big Five
Five dimensions) for the prediction of task performance criteria, 14- l 7 for
as its a priori conceptual basis). Four distinct measures were identified in
job dedication criteria, and 19-23 for criteria fitting the definition of
the studies collected for the present review: the NEO Personality Inventory
interpersonal facilitation.
(NEO-PI), including the revised (NEO-PI-R) and five-factor inventory
(NEO-FFI) versions (Costa & McCrae, 1992), Goldberg's Big Five mark-
ers (Goldberg, 1992), the Hogan Personality Inventory (HPI; Hogan & Computation o f Validity Coefficients
Hogan, 1995), and the Personal Characteristics Inventory (PCI; Barrick &
Mount, 1993). Finally, the study had to include an explicit measure of job Within individual studies, there were instances in which correlation
performance or training performance as the criterion of interest. coefficients from a single sample had to be combined. On the predictor
side, for the HPI, some studies reported separate correlations for the
Coding o f Potential Moderators and Study Characteristics Ambition and Sociability subscales of the Extraversion dimension and tbr
the Intellectance and School Success subscales of the Opennessfintellect
Consistent with previous meta-analyses (Barrick & Mount, 1991; Sal- dimension, rather than correlations at the dimension level. In these cases,
gado, 1997), two study characteristics were coded and treated as potential rather than averaging across the subscales, we computed the composite
moderators of the relations between the Big Five and job performance: type score correlation (see Hunter & Schmidt, 1990, pp. 454-463) to estimate
of worker occupation and type of performance criterion. Each of the the correlation between the sum of the two lower level subscales and job
potential moderator variables was independently coded by both Gregory performance. When correlations between the lower level subscales were
M. Hurtz and John J. Donovan to ensure accuracy and completeness of not provided in a study, we entered the correlations presented in the HPI
coding. Overall, a high degree of initial agreement (98%) was obtained manual into the composite score correlation formula. Combining the lower
between the two independent raters, and divergent ratings were discussed level scales in this manner is entirely consistent with the tact that these
by the authors until there was an agreement about the proper coding of the subscales were derived directly from the dimension-level scales; thus, their
study in question. sum directly assesses the dimension-level construct. Therefore, this does
Worker occupation. The first characteristic coded for was the occupa- not undermine our purpose of including only explicit Big Five measures.
tion of the workers being examined in the study. A four-category classi- Similarly on the criterion side, some studies provided correlations be-
fication scheme was used to identify the occupation of all research partic- tween the Big Five and separate dimensions of job performance without
ipants: sales workers, customer service representatives, managers, and providing a correlation with the composite criterion score. In these cases,
skilled and semiskilled workers. Approximately 22% (10 of 45) of the we again estimated the composite score correlation rather than simply
validity coefficients included in the present review came from studies averaging across performance dimensions. When the correlations between
examining sales jobs, 27% (12 of 45) came from customer service jobs, 9% performance dimensions were not provided, we entered .55 into the com-
(4 of 45) were based on managerial jobs, and 31% (14 of 45) came from posite score formula. We derived this estimate by first computing the
skilled and semiskilled jobs. Approximately 11% (5 of 45) were not average correlation among dimensions within the studies that did provide
 BIG FIVE AND JOB PERFORMANCE 873

such information and then computing the mean sample-size weighted Results
correlation across these studies. For the separate analyses of task and
contextual performance dimensions, we used the same procedure for com- Overall Validity Coefficients
bining correlations from a single sample that were based on multiple rating
scales classified into a common dimension. T a b l e 1 p r e s e n t s the r e s u l t s o f t h e o m n i b u s m e t a - a n a l y s i s
When conducting the actual meta-analysis, we used the Hunter-Schmidt a c r o s s o c c u p a t i o n s a n d p e r f o r m a n c e criteria. T h e s e a n a l y s e s
validity generalization framework (Hunter & Schmidt, 1990). Using this w e r e b a s e d on a r a n g e o f 3 5 - 4 5 c o r r e l a t i o n s a n d 5 , 5 2 5 - 8 , 0 8 3
framework, we obtained the mean sample size-weighted correlations, the job applicants and incumbents. The mean sample-size weighted
estimated true or operational validities corrected for sampling error, range c o r r e l a t i o n s (?) r a n g e d f r o m .04 to .14 a c r o s s d i m e n s i o n s a n d
restriction, criterion unreliability, and the estimated true-score correlations
are s u b s t a n t i a l l y l o w e r t h a n the m e a n c o r r e l a t i o n o f .21 f o u n d
with additional corrections for predictor unreliability.
by S c h m i t t et al. (1984) a n d v e r y s i m i l a r to t h o s e f o u n d b y
As in the previous meta-analyses we have just reviewed, these correc-
B a r r i c k a n d M o u n t (1991; r a n g i n g . 0 3 - . 13) a n d S a l g a d o (1997;
tions had to be made by way of artifact distributions because of a low rate
of reporting the statistics that are necessary for applying corrections to the r a n g i n g . 0 1 - . 10). T h e e s t i m a t e d true v a l i d i t i e s (Pv) for explicit
individual coefficients. Two artifact distributions were created for the m e a s u r e s o f the B i g F i v e r a n g e d f r o m .06 to .20, a n d the
criterion reliabilities. For analyses in which only subjective ratings of e s t i m a t e d t r u e - s c o r e c o r r e l a t i o n s (Pc) r a n g e d f r o m .07 to .22.
performance were involved, we created a distribution by augmenting the C o n s i s t e n t with B a r r i c k a n d M o u n t (1991) a n d S a l g a d o (1997),
few interrater reliability coefficients obtained from our sample of studies the h i g h e s t validity o f the B i g F i v e d i m e n s i o n s w a s that for
with those presented in Rothstein (1990). This distribution had a mean C o n s c i e n t i o u s n e s s (Pv = .20), w h i c h d e m o n s t r a t e d a low to
criterion reliability of .53 (SD = .15). For those analyses in which a m o d e r a t e level o f validity. T h e 9 0 % credibility interval for this
combination of objective and subjective criteria were used, we added to the
d i m e n s i o n did not i n c l u d e zero, s u g g e s t i n g t h e a b s e n c e o f
previous distribution the reliabilities presented in our sample for objective
m o d e r a t o r s in this e s t i m a t e o f the true v a l i d i t y ( H u n t e r &
criteria and those presented by Hunter, Schmidt, and Judeisch (1990).
Adding these reliabilities created a distribution with a mean of .59 (SD = S c h m i d t , 1990; W h i t e n e r , 1990). E m o t i o n a l Stability also h a d a
.19). Although this combined distribution is weighted rather heavily with credibility interval that w a s g r e a t e r t h a n zero, a l t h o u g h its
subjective ratings, this is entirely consistent with the fact that approxi- e s t i m a t e d true validity w a s s u b s t a n t i a l l y l o w e r (p~ = .13).
mately 90% of the criteria in our sample of studies were subjective in
nature.
For corrections for predictor unreliability, we created separate artifact
Validity Coefficients by Occupation
distributions for each of the Big Five dimensions by augmenting the Table 2 presents the results o f the m o d e r a t o r analysis for the
reliability estimates provided in our sample of studies with those from the
occupational categories. Despite the lack o f m o d e r a t o r s indicated
inventory manuals. This provided distributions with mean predictor reli-
by the credibility intervals for C o n s c i e n t i o u s n e s s and E m o t i o n a l
abilities ranging from .76 (SD = .08; Agreeableness) to .86 (SD = .04;
Emotional Stability). For range restriction corrections, we found very few Stability in the o m n i b u s analysis, we c a r d e d out all m o d e r a t o r
unrestricted standard deviations reported in the studies for computing the u a n a l y s e s for each o f the Big Five for the sake o f c o m p a r i s o n . For
values. Thus, we used two strategies for obtaining unrestricted standard all four o f the occupational categories, C o n s c i e n t i o u s n e s s exhib-
deviations. First, we attempted to contact the authors of the inventories to ited the h i g h e s t e s t i m a t e d true validity. It is interesting to note that
obtain standard deviations from unrestricted samples of applicants. Second, despite the indication of no m o d e r a t o r s for C o n s c i e n t i o u s n e s s , the
following Salgado's (1997) strategy, we used standard deviations provided e s t i m a t e d true validity for this d i m e n s i o n ranged f r o m .15 to .26
in the inventory manuals as the unrestricted values. As we did not have across occupations. Its h i g h e s t validities were for sales (p~ = .26)
enough information to create reliable separate distributions for each of the
and c u s t o m e r service (Pv = .25) jobs. T h e m a g n i t u d e s o f t h e s e
five dimensions, we created a single artifact distribution for use in all our
validities are moderate, and those for the r e m a i n i n g Big Five
analyses. This distribution of u values had a mean of .92 (SD = .27).
d i m e n s i o n s r e m a i n e d low across all occupations.
Overall, our artifact distributions were very similar to those used in the
previous meta-analyses. Corrections based on these distributions were It is noteworthy, however, that s o m e o f the low validities for the
conducted interactively using software described by Hunter and Schmidt other Big Five d i m e n s i o n s appear to be rather stable, in that their
(1990), on the basis of the recommendations of Law, Schmidt, and Hunter credibility intervals fall above zero. For sales jobs, E m o t i o n a l
(1994). Stability (Pv = -13) a n d Extraversion (p~ = .15) appear to h a v e

Table 1
Overall Validity Coefficients by Personality Dimension

Big Five dimension k N ? S~ S~ S2. . . . SLs % VE Pc Pv SDo,. 90% CV

Conscientiousness 45 8,083 .14 .0161 .0054 .0016 .0091 44 .22 .20 .14 .03
Emotional Stability 37 5,671 .09 .0084 .0065 .0007 .0013 85 .14 .13 .05 .06
Agreeableness 40 6,447 .07 .0108 .0062 .0005 .0041 62 .13 .11 .09 - .01
Extraversion 39 6,453 .06 .0111 .0060 .0004 .0047 57 .10 .09 .10 - .04
Openness to Experience 35 5,525 .04 .0093 .0064 .0002 .0028 70 .07 .06 .08 -.04

Note. k = number of validity coefficients; N = total sample size; g = sample-size weighted mean observed validity; S~ = total observed variance in ~;
S~ = variance due to sampling error; S21eas = variance due to measurement artifacts; S~es = residual variance; % VE = percentage of variance accounted
for by sampling error and measurement artifacts; Pc = true-score correlation; Pv = true (operational) validity; SDpv = standard deviation of true validity;
CV = credibility value (lower bound of credibility interval for Or).
874 HURTZ AND DONOVAN

Table 2
Validity Coefficientsfor Personality Dimensions by Occupational Category
Big Five dimension k N ? S2r Be2 S2 are
s" % VE Pc Pv SDo,, 90% CV

Sales
Conscientiousness 10 1,369 .18 .0117 .0069 .0026 .0021 82 .29 .26 .07 .17
Emotional Stability 7 799 .09 .0082 .0087 .0007 .0000 115 .15 .13 .00 .13
Agreeableness 8 959 .03 .0098 .0084 .0001 .0013 87 .06 .05 .05 -.02
Extraversion 8 1,044 .10 .0117 .0076 .0009 .0033 72 .16 .15 .08 .04
Openness to Experience 6 732 .03 .0150 .0083 .0001 .0067 55 .04 .04 .12 -.12

Customer service
Conscientiousness 12 1,849 .17 .0121 .0062 .0023 .0036 70 .27 .25 .09 .13
Emotional Stability 10 1,614 .08 .0052 .0062 .0006 .0000 129 .13 .12 .00 .12
Agreeableness 11 1,719 .11 .0038 .0063 .0011 .0000 193 .19 .17 .00 .17
Extraversion 10 1,640 .07 .0117 .0061 .0004 .0052 56 .11 .11 .11 -.03
Openness to Experience 9 1,535 .10 .0043 .0058 .0010 .0000 158 .17 .15 .00 .15

Managers
Conscientiousness 4 495 .11 .0451 .0079 .00ll .0361 20 .19 .17 .28 -.19
Emotional Stability 4 495 .08 .0088 .0080 .0006 .0002 98 .13 .12 .02 .10
Agreeableness 4 495 - .03 .0040 .0081 .0001 .0000 205 - .04 - .04 .00 - .04
Extraversion 4 495 .08 .0045 .0080 .0006 .0000 192 .13 .12 .00 .12
Openness to Experience 4 495 -.02 .0111 .0081 .0000 .0029 74 - .03 -.03 .08 - . 13

Skilled and semiskilled

Conscientiousness 14 3,481 .10 .0147 .0040 .0009 .0098 33 .17 .15 .15 -.03
Emotional Stability 11 1,874 .06 .0110 .0059 .0003 .0048 56 .09 .08 .10 -.05
Agreeableness 12 2,385 .06 .0103 .0050 .0004 .0049 52 .11 .10 .10 - .04
Extraversion 12 2,385 .00 .0080 .0051 .0000 .0029 63 .01 .01 .08 - . 10
Openness to Experience 11 1,874 -.01 .0062 .0059 .0000 .0003 95 -.02 -.02 .03 -.05

Note. k = number of validity coefficients; N = total sample size; ? = sample-size weighted mean observed validity; S2 = total observed variance in ?;
S~ = variance due to sampling error; sZea~ = variance due to measurement artifacts; S2e~ = residual variance; % VE = percentage of variance accounted
for by sampling error and measurement artifacts; Pc = true-score correlation; Pv = true (operational) validity; SDov = standard deviation of true validity:
CV = credibility value (lower bound of credibility interval for Pv).

low but stable true validities. This same general pattern emerged although Extraversion (Pv = -17) and Agreeableness (Pv = • 18)
for managerial jobs, although the small number o f studies (k = 4) had the highest validities.
located for estimating this true validity may render this finding Table 4 shows the separate analyses o f the Big Five as predic-
tenuous. Customer service jobs appear more complex in that tors o f task performance, j o b dedication, and interpersonal facili-
Emotional Stability (Pv = .12), Agreeableness (Pv = .17), and tation. Recent research and theory explicating the dimensionality
Openness to Experience (Pv = -15) exhibited rather low but stable o f the j o b performance domain has suggested that personality
true validities. This may indicate a s o m e w h a t more complex pat- should predict the contextual performance dimensions o f job ded-
tern o f relationships between personality and performance in jobs ication and interpersonal facilitation more strongly than task per-
that involve interpersonal interactions than is captured solely by formance does (Borman & Motowidlo, 1993, 1997; Motowidlo &
assessing Conscientiousness. In contrast, the true validity esti- Van Scotter, 1994; Van Scotter & Motowidlo, 1996). Our analyses
mates for skilled and semiskilled jobs, which may often involve a
show that Conscientiousness predicted all three criteria with ap-
smaller interpersonal c o m p o n e n t o f performance, tended to be
proximately the same level o f true validity (Pv = .15-. 18), al-
rather small across all of the Big Five, and these validities appear
though the credibility intervals indicate that this true validity was
rather unstable in light of their credibility intervals.
only stable for the interpersonal facilitation criterion. Emotional
Stability appeared to have a low but very stable true validity across
Validity Coefficients by Criterion Type these three criteria (Pv = • 13-. 16). For the interpersonal facilitation
Table 3 presents the results o f the moderator analysis for the criterion, Agreeableness (Pv = • 17) rivaled both Conscientiousness
separate predictions o f job proficiency and training proficiency. (Pv = .16) and Emotional Stability (Pv = • 16) in its estimated true
For j o b proficiency, virtually the same pattern and magnitude o f validity. This supports Van Scotter and M o t o w i d l o ' s finding that
validities emerged as was found in the omnibus analysis, which is although Agreeableness does not influence task performance, it
not surprising given the fact that over 90% o f the individual does appear to influence ratings o f interpersonal facilitation. It
correlations across dimensions involved job proficiency criteria. should be noted, however, that none o f these analyses for the task
The small number o f correlations summarized for training profi- and contextual performance criteria revealed stronger true validi-
ciency renders interpretation o f the true validity estimates tenuous, ties than did the overall performance analyses.
 BIG FIVE AND JOB PERFORMANCE 875

Table 3
Validity Coefficients for Personality Dimensions by Criterion Type
2 2
Big Five dimension k N ~ S2 S~ Sme~ S~.~ % VE pc p~ SDov 90% CV

Job performance
Conscientiousness 42 7,342 .15 .0148 .0055 .0019 .0074 50 .24 .22 .13 .06
Emotional Stability 35 5,027 .09 .0089 .0069 .0007 .0013 85 .15 .14 .05 .07
Agreeableness 38 5,803 .07 .0111 .0065 .0004 .0042 62 .12 .10 .10 -.02
Extraversion 37 5,809 .06 .0118 .0064 .0003 .0051 57 .09 .09 .11 -.05
Openness to Experience 33 4,881 .03 .0097 .0068 .0001 .0028 71 .06 .05 .08 -.05

Training performance
Conscientiousness 3 741 .02 .0145 .0041 .0000 .0104 28 .03 .03 .15 -.16
Emotional Stability 2 644 .06 .0030 .0031 .0003 .0000 111 .09 .08 .00 .08
Agreeableness 2 644 .12 .0049 .0030 .0013 .0006 88 .21 .18 .04 .13
Extraversion 2 644 .12 .0020 .0030 .0012 .0000 207 .19 .17 .00 .17
Openness to Experience 2 644 .08 .0042 .0031 .0007 .0005 88 .14 .13 .03 .08

Note. k = number of validity coefficients; N = total sample size; ~ = sample-size weighted mean observed validity; S~ = total observed variance in ~;
S~ = variance due to sampling error; SZmea~= variance due to measurement artifacts; S ~ = residual variance; % VE = percentage of variance accounted
for by sampling error and measurement artifacts; pc = true-score correlation; p~ = true (operational) validity; SDo~ = standard deviation of true validity;
CV = credibility value (lower bound of credibility interval for P0.

Discussion the h i g h e s t validity o f the B i g Five d i m e n s i o n s for overall j o b

p e r f o r m a n c e . Furthermore, our e s t i m a t e d true-score correlation o f
T h e m a i n p u r p o s e o f this study w a s to provide a c o n f i r m a t o r y .22 (and true validity o f .20) w a s virtually identical in m a g n i t u d e
m e t a - a n a l y s i s o f the relation b e t w e e n the Big Five and j o b perfor- to Barrick a n d M o u n t ' s estimate. T h i s f i n d i n g alleviates c o n c e r n
m a n c e by i n c l u d i n g only scales that were explicitly d e s i g n e d to that Barrick and M o u n t ' s h e a v i l y cited results u n d e r e s t i m a t e d the
m e a s u r e the Big Five personality d i m e n s i o n s . O u r overall results overall true validity o f the C o n s c i e n t i o u s n e s s d i m e n s i o n as a result
were h i g h l y consistent with the original w o r k o f Barrick and o f their categorization procedures. O n the other hand, our f i n d i n g s
M o u n t (1991), in that C o n s c i e n t i o u s n e s s w a s again f o u n d to h a v e indicate that at least for single-scale, global B i g Five m e a s u r e s , the

Table 4
Validity Coefficients for Personality Dimensions by Criterion Dimension

Big Five dimension k N ~ S~ S~ S~ea~ Sr%s % VE Pc Pv SDpv 90% CV

Task performance
Conscientiousness 12 2,197 .10 .0138 .0054 .0008 .0076 45 .16 .15 .13 -.02
Emotional Stability 8 1,243 .09 .0015 .0064 .0007 .0000 463 .14 .13 .00 .13
Agreeableness 9 1,754 .05 .0090 .0051 .0002 .0037 59 .08 .07 .09 - .05
Extraversion 9 1,839 .04 .0052 .0049 .0002 .0001 98 .07 .06 .02 .04
Openness to Experience 7 1,176 -.01 .0237 .0060 .0000 .0177 25 -.01 -.01 .20 -.26

Job dedication
Conscientiousness 17 3,197 .l2 .0203 .0052 .0013 .0139 32 .20 .18 .17 - .04
Emotional Stability 15 2,581 .09 .0059 .0058 .0007 .0000 109 .14 .13 .00 .13
Agreeableness 17 3,197 .06 .0096 .0053 .0003 .0040 59 .10 .08 .09 -.03
Extraversion 16 3,130 .03 .0111 .0051 .0001 .0059 47 .05 .05 .11 -.10
Openness to Experience 14 2,514 .01 .0108 .0056 .0000 .0052 52 .01 .01 .11 -.13

Interpersonal facilitation
Conscientiousness 23 4,301 .11 .0083 .0053 .0010 .0020 76 .18 .16 .07 .07
Emotional Stability 21 3,685 .10 .0046 .0056 .0010 .0000 142 .17 .16 .00 .16
Agreeableness 23 4,301 .11 .0117 .0052 .0012 .0053 55 .20 .17 .11 .03
Extraversion 21 4,155 .06 .0105 .0050 .0004 .0051 52 .11 .10 .11 -.04
Openness to Experience 19 3,539 .03 .0075 .0054 .0001 .0020 73 .05 .05 .07 -.04

Note. k = number of validity coefficients; N = total sample size; ~ = sample-size weighted mean observed validity; Sr2- = total observed variance in f;
S~2 = variance due to sampling error; S~mea~= variance due to measurement artifacts; S~s = residual variance; % VE = percentage of variance accounted
for by sampling error and measurement artifacts; Pc = true-score correlation; Pv = true (operational) validity; SDpv = standard deviation of true validity;
CV = credibility value (lower bound of credibility interval for pv).
876 HURTZ AND DONOVAN

validity estimates for Conscientiousness provided by Mount and What degree of utility do these global Big Five measures offer
B arrick (1995 ) and S algado (1997 ) appear to be overestimates. We for predicting job performance? Overall, it appears that global
offer from our results an estimated true criterion-related validity of measures of Conscientiousness can be expected to consistently add
.20 for actual Big Five measures of Conscientiousness. a small portion of explained variance in job performance across
It is also noteworthy that Emotional Stability shows rather jobs and across criterion dimensions. In addition, for certain jobs
consistent (although low) levels of criterion-related validity. In and for certain criterion dimensions, certain other Big Five dimen-
addition, the separate analyses for the different occupational cat- sions will likely add a very small but consistent degree of ex-
egories provide a more complex picture of the validities of the Big plained variance. If the global Big Five measure is uncorrelated
Five than do prior reviews, in that the dimensions beyond Con- with the other predictors that are currently used ~br a job (e.g.,
scientiousness begin to show low but rather stable validities for personality tends to be uncorrelated with cognitive ability; Day &
certain occupations. In particular, for jobs involving customer Silverman, 1989; Rosse, Miller, & Barnes, 1991), then even this
service, Agreeableness, Openness to Experience, and Emotional small incremental explained variance can, under certain circum-
Stability had low levels of validity (pvs ranging. 12-. 17) but zero stances, make a practically significant contribution to predictive
residual variance in the population estimate after the effects of efficiency for a job and perhaps contribute to a reduction in
sampling error and measurement artifacts were removed. Similarly adverse impact (Hattrup, Rock, & Scalia, 1997; Murphy &
for sales and perhaps for managerial jobs, Emotional Stability and Shiarella, 1997; Schmitt, Rogers, Chan, Sheppard, & Jennings,
Extraversion had rather low but stable validities. A common theme 1997; but see Ryan, Ployhart, & Friedel, 1998).
running through customer service, sales, and managerial jobs that However, if the relevant aspects of the applicant's personality
differentiates them from skilled and semiskilled jobs is the inter- are already partially captured through other selection techniques
personal component of performing these jobs; this probably ac- such as reference checks and interviews, the small potential con-
counts for the more stable validities of these personality dimen- tribution of a global Big Five measure will likely diminish. In
sions for these types of jobs. addition, the potentially negative impact of faking on the utility of
personality measures (Christiansen, Goffin, Johnston, & Rothstein,
The results of partitioning the criterion domain into task perfor-
1994; Ellingson, Sackett, & Hough, 1999; Rosse, Stecher, Miller,
mance, job dedication, and interpersonal facilitation shed further
& Levin, 1998) and the potentially negative applicant reactions
light on these issues. Contrary to our expectations, our analyses
(Rosse, Miller, & Stecher, 1994; Steiner & Gilliland, 1996) also
showed that partitioning the criterion domain in this manner did
raise the question of whether this small addition of explained
not bring about stronger criterion-related validities in comparison
variance should be interpreted as enthusiastically as it recently has
with analyzing a general job proficiency category. Conscientious-
been. Thus, although the low to moderate validity coefficient for
ness predicted all three performance dimensions equally well
Conscientiousness tends to generalize quite well across occupa-
(Pv = .15-.18), and the same was found for Emotional Stability
tions and job performance criteria and although other Big Five
(Or = •13-. 16). However, Agreeableness did emerge as a poten-
dimensions appear to have meaningful relations with certain cri-
tially valid predictor, predicting interpersonal facilitation as
teria or for certain jobs, we do not see evidence that Conscien-
strongly as did Conscientiousness and Emotional Stability. When
tiousness or any of the other Big Five dimensions should be
one considers the validities and credibility intervals together, Con-
granted a status similar to that of general cognitive ability for
scientiousness, Emotional Stability, and Agreeableness appear to personnel selection purposes (cf. Behling, 1998).
have a rather stable impact on the interpersonal facilitation criteria, In terms of theory rather than practice, however, we do interpret
suggesting that perhaps this performance dimension is influenced our findings as indicating a pattern of theoretically meaningful
in a consistent manner by certain personality traits. This is con- relations between the broad personality dimensions and job per-
sistent with the previous suggestion that personality may have a formance that should be explored in future research, perhaps using
more stable impact on jobs that are more interpersonal in nature facet scales of the Big Five dimensions. Although the strength of
(e.g., customer service, sales, management). The same statement the relations are low to moderate, different personality dimensions
may be true of job performance dimensions that are more inter- appear to affect performance in different types of jobs or along
personal in nature. different dimensions. In a relative sense, the Conscientiousness
Although these findings shed some light on the potential impact dimension does appear to have the strongest relation to overall job
of personality variables on dimensions of job performance, we performance. People who describe themselves as hard-working,
now return to the issue of the absolute magnitudes of the estimated reliable, organized, and so on do appear to perform somewhat
true validity coefficients for these variables and the implications of better than do those who believe they are less strong in these
these validities for the utility of the Big Five for personnel selec- characteristics. It is also interesting that Emotional Stability
tion. In general, our analyses suggest that the validities of the Big showed a rather stable influence on performance throughout nearly
Five, including Conscientiousness, tend to be low to moderate in all of our analyses. It appears that being calm, secure, well-
magnitude. One of the major implications of this meta-analysis, adjusted, and low in anxiety has a small but consistent impact on
then, is that stating that Conscientiousness is a valid predictor of job performance. Agreeableness also gains importance for those
job performance paints an inaccurate image of the true validity of jobs that require interpersonal interactions, so that being likeable,
these global Big Five measures in an absolute sense. We suggest cooperative, and good-natured has a small but consistent impact on
that the estimated true validity of .20 for the global Conscientious- performance. Finally, being Extraverted appears to influence sales
ness dimension is not as impressive as one would expect, given the and perhaps managerial jobs, and Openness to Experience appears
recent enthusiasm surrounding its use as a predictor of job to affect performance in customer service jobs. Although these
performance. theoretically meaningful relations are rather low in magnitude at
 BIG FIVE AND JOB PERFORMANCE 877

the broad dimension level of the Big Five, the magnitude of these tently had equivalent or higher levels of criterion-related validity
correlations might be enhanced if the most relevant specific facets in comparison with employees' self-reports. Although the practice
of these broad dimensions could be specified. of using rating sources other than oneself is not likely to be
We suggest, then, that the Big Five framework and the patterns adopted in personnel selection practice, such alternative measure-
of small to moderate validities for these broad dimensions that ment methods could help gain a better understanding of the aspects
have begun to emerge should be used in future research to help of personality that affect performance.
guide the selection back "downward" toward somewhat narrower
personality facets with theoretical links to the performance dimen-
sions under investigation. If a broad, global performance criterion Limitations
is of interest, perhaps a global Conscientiousness scale will suffice
At least two limitations of the current meta-analysis should be
with a moderate level of validity. However, if multiple perfor-
pointed out. First, several of our moderator analyses were based on
mance dimensions such as those distinguishing task performance
a relatively small number of correlations, especially for the man-
from contextual performance, or perhaps those consistent with
agerial occupation category and for training proficiency. This
other typologies such as that presented by Campbell (1990), will
renders any conclusions based on these moderator categories ten-
be delineated, then perhaps narrower facets of performance with
uous. It is especially unfortunate that so little research has been
strong theoretical links to those criteria can be identified and used
published using managers, because clarification on the impact of
individually or in combination to enhance their criterion-related
the Big Five for this occupation would be beneficial for selection
validity.
research and practice.
We also note that the formation of optimal composites may
Second, our categorization of job performance dimensions into
involve grouping facets from across the five broad dimensions. For
task performance, job dedication, and interpersonal facilitation
example, combining selected facets of Conscientiousness, Emo-
could be criticized on the same grounds that we used to criticize
tional Stability, and Agreeableness may optimize the prediction of
the earlier meta-analyses for their categorization of personality
an interpersonal facilitation criterion. The circumplex models of
measures into the Big Five. Perhaps as the body of research on
the Big Five presented by Hofstee, de Raad, and Goldberg (1992)
dimensions of job performance develops, an assessment of the
and Johnson and Ostendorf (1993) could also prove useful in this
predictability of these or other performance dimensions using a
regard by guiding the formation of predictor scales that simulta-
priori measures of the relevant dimensions can be undertaken so as
neously represent aspects of two dimensions, in a sense falling
to avoid any problems with this type of classification procedure.
between two of the broader dimensions. This is a hypothesis that
Given that we found few effects of such potential for classification
deserves consideration in future research.
errors on the resulting validities for the Big Five, we may be able
In addition to further exploration of more specific links between
to predict that such an analysis would yield results that differ little
dimensions of personality and job performance, research that more
from those we obtained in this study. However, this is an empirical
completely delineates the nomological network connecting person-
question that would need to be addressed directly in future
ality to job performance is needed. Much of the research to date
research.
has taken a very practical perspective, focusing on the bivariate
correlation between personality and performance. However, if we
are to truly understand the relationship between personality and Conclusions
job performance, we must move beyond this bivariate relationship
and toward specifying the intervening variables that link these In summary, the present meta-analysis provides a review of the
domains. The Conscientiousness trait, for example, is often dis- criterion-related validities of the Big Five personality dimensions,
cussed in a manner that assumes it has motivational implications. as measured by scales that were developed explicitly according to
Motivational variables, then, should be examined more extensively the five-factor model. Although we have interpreted this evidence
as intervening variables in a multivariate model. Some research of the criterion-related validity of Conscientiousness somewhat
has, in fact, found Conscientiousness to influence performance less optimistically than many researchers have tended to do in
through its effects on such motivational variables as performance recent years, we nevertheless suggest that the potential exists for
expectancies, self-efficacy, and goal setting (Barrick, Mount, & improving the validity of personality predictors. We encourage
Strauss, 1993; Gellatly, 1996; Martocchio & Judge, 1997). Better future research aimed at theory-based matching of personality
explication of this nomological network for different conceptual- constructs and dimensions of job performance, perhaps using
izations of the personality domain (e.g., narrower facets or Big composites of narrower Big Five facets. We also encourage re-
Five dimensions other than Conscientiousness) and for different search aimed at building a more extensive multivariate model of
dimensions of job performance may aid in a better understanding the personality-job performance relation.
of how personality affects job performance. With better identifi- We conclude that global measures of the Conscientiousness
cation of intervening variables, the total effects of personality on dimension have a rather moderate impact on performance, al-
job performance may emerge more strongly than the simple biva- though this validity does appear rather stable and generalizable
riate correlation coefficient has demonstrated. across occupations and criteria. Although they are less generaliz-
Finally, for the sake of understanding the impact of personality able, we also conclude that personality traits other than Conscien-
on job performance, it would also be interesting to explore these tiousness are nearly equally important for certain occupations and
relations using alternative measurement methods. Mount, Barrick, criteria. Our hope is that the results of this review encourage
and Strauss (1994), for example, presented some evidence that realistic expectations about the potential contribution of Consci-
supervisor, coworker, and customer ratings of personality consis- entiousness measures to selection utility and encourage further
878 HURTZ AND DONOVAN

exploration o f the impact o f personality variables on j o b perfor- Gellatly, I. R. (1996). Conscientiousness and task performance: Test of a
mance beyond the global Conscientiousness dimension. cognitive process model. Journal of Applied Psychology, 81, 474-482.
Goldberg, L. R. (1992). The development of markers of the big-five factor
structure. Psychological Assessment, 4, 26-42.
References Goldberg, L. R. (1993). The structure of phenotypic personality traits.
American Psychologist, 48, 26-34.
References marked with an asterisk indicate studies that con- Guion, R. M., & Gottier, R. F. (1965). Validity of personality measures in
tributed a validity coefficient to the meta-analysis. personnel selection. Personnel Psychology, 18, 135-164.
Hattrup, K., Rock, J., & Scalia, C. (1997). The effects of varying concep-
Barrick, M. R., & Mount, M. K. (1991). The Big Five personality dimen-
tualizations of job performance on adverse impact, minority hiring, and
sions and job performance: A meta-analysis. Personnel Psychology, 44,
predicted performance. Journal of Applied Psychology, 82, 656-664.
1-26.
*Hayes, T. L., Roehm, H. A., & Castellano, J. P. (1994). Personality
*Barrick, M. R., & Mount, M. K. (1993). Autonomy as a moderator of the
correlates of success in total quality manufacturing. Journal of Business
relationships between the Big Five personality dimensions and job
and Psychology, 8, 397-411.
performance. Journal of Applied Psychology, 78, 111-118.
Hofstee, W. K. B., de Raad, B., & Goldberg, L. R. (1992). Integration of
*Barrick, M. R., & Mount, M. K. (1996). Effects of impression manage-
the Big Five and circumplex approaches to trait structure. Journal of
ment and self-deception on the predictive validity of personality con-
Personality and Social Psychology, 63, 146-163.
structs. Journal of Applied Psychology, 81, 261-272.
Hogan, R., & Hogan, J. (1995). Hogan Personality Inventory manual.
*Barrick, M. R., Mount, M. K., & Strauss, J. P. (1993). Conscientiousness
Tulsa, OK: Hogan Assessment Systems.
and performance of sales representatives: Test of the mediating effects
Hogan, R., Hogan, J., & Roberts, B. W. (1996). Personality measurement
of goal setting. Journal of Applied Psychology, 78, 715-722.
and employment decisions: Questions and answers. American Psychol-
*Barrick, M. R., Mount, M. K., & Strauss, J. P. (1994). Antecedents of
ogist, 51, 469-477.
involuntary turnover due to a reduction in force. Personnel Psychol-
Hogan, J., & Ones, D. S. (1997). Conscientiousness and integrity at work.
ogy, 47, 515-535.
In R. Hogan, J. Johnson, & S. Briggs (Eds.), Handbook of personality
Behling, O. (1998). Employee selection: Will intelligence and conscien-
psychology (pp. 849-870). San Diego, CA: Academic Press.
tiousness do the job? Academy of Management Executive, 12, 77-86.
*Hogan, J., Rybicki, S. L., Motowidlo, S. J., & Borman, W. C. (1998).
Borman, W. C., & Motowidlo, S. J. (1993). Expanding the criterion
Relations between contextual performance, personality, and occupa-
domain to include elements of contextual performance. In N. Schmitt &
tional advancement. Human Performance, 11, 189-207.
W. C. Borman (Eds.), Personnel selection in organizations (pp. 71-98).
Hough, L. M. (1992). The "Big Five" personality variables-construct
San Francisco: Jossey-Bass.
Borman, W. C., & Motowidlo, S. J. (1997). Task performance and con- confusion: Description versus prediction. Human Performance, 5, 139-
textual performance: The meaning for personnel selection research. 155.
Human Performance, 10, 99-109. Hunter, J. E., & Schmidt, F. L. (1990). Methods ofmeta-analysis: Cor-
Campbell, J. P. (1990). Modeling the performance prediction problem in recting error and bias in research findings. Newbury Park, CA: Sage.
industrial and organizational psychology. In M. D. Dunnette & L. M. Hunter, J. E., Schmidt, F. L., & Judiesch, M. K. (1990). Individual
Hough (Eds.), Handbook of industrial and organizational psychology differences in output variability as a function of job complexity. Journal
(2nd ed., Vol. 1, pp. 687-732). Palo Alto, CA: Consulting Psychologists of Applied Psychology, 75, 28-43.
Press. Iaffaldano, M. T., & Muchinsky, P. M. (1985). Job satisfaction and job
*Cellar, D. F., DeGrendel, D. J. D., Klawsky, J. D., & Miller, M. L. (1996). performance: A meta-analysis. Psychological Bulletin, 97, 251-273.
The validity of personality, service orientation, and reading comprehen- *Jacobs, R. R., Conte, J. M., Day, D. V., Silva, J. M., & Harris, R. (1996).
sion measures as predictors of flight attendant training performance. Selecting bus drivers: Multiple predictors, multiple perspectives on
Journal of Business and Psychology, 11, 43-55. validity, and multiple estimates of utility. Human Performance, 9, 199-
*Cellar, D. F., Miller, M. L., Doverspike, D. D., & Klawsky, J. D. (1996). 217.
Comparison of factor structures and criterion related validity coefficients John, O. P. (1990). The "Big Five" factor taxonomy: Dimensions of
for two measures of personality based on the five factor model. Journal personality in the natural language and in questionnaires. In L. A. Pervin
of Applied Psychology, 81, 694-704. (Ed.), Handbook of personality: Theory and research (pp. 66-100).
Christiansen, N. D., Goffin, R. D., Johnston, N. G., & Rothstein, M. G. New York: Guilford.
(1994). Correcting the 16PF for faking: Effects on criterion-related Johnson, J. A., & Ostendorf, F. (1993). Clarification of the five-factor
validity and individual hiring decisions. Personnel Psychology, 47, model with the abridged Big Five dimensional circumplex. Journal of
847- 860. Personality and Social Psychology, 65, 563-576.
Cohen, J. (1988). Statistical power analysis for the behavioral sciences *Kolz, A. R., Cardillo, E. P., & Pena, S. A. (1998, April). Cognitive ability
(2nd ed.). Hillsdale, NJ: Erlbaum. and personality as predictors of performance and counterproductive
Costa, P. T., Jr., & McCrae, R. R. (1992). Revised NEO Personality behavior. Paper presented at the annual conference of the Society for
Inventory (NEO-P1-R) and NEO Five-Factor Inventory (NEO-FF1) pro- Industrial and Organizational Psychology, Dallas, TX.
fessional manual. Odessa, FL: Psychological Assessment Resources. Law, K. S., Schmidt, F. L., & Hunter, J. E. (1994). A test of two
*Crant, J. M. (1995). The Proactive Personality Scale and objective job refinements in procedures for meta-analysis. Journal c~"Applied Psy-
performance among real estate agents. Journal of Applied Psychol- chology, 79, 978-986.
ogy, 80, 532-537. *Lyne, R., Sinclair, R. R., & Gerhold, C. (1997, April). Personality and job
Day, D. V., & Silverman, S. B. (1989). Personality and job performance: performance: Matching predictor and criterion domains. Paper pre-
Evidence of incremental validity. Personnel Psychology, 42, 25-36. sented at the annual conference of the Society for Industrial and Orga-
Digman, J. M. (1990). Personality structure: Emergence of the five-factor nizational Psychology, St. Louis, MO.
model. Annual Review of Psychology, 41, 417- 440. *Mabon, H. (1998). Utility aspects of personality and performance. Human
Ellingson, J. E., Sackett, P. R., & Hough, L. M. (1999). Social desirability Performance, 11, 289-304.
corrections in personality measurement: Issues of applicant comparison *Martocchio, J. J., & Judge, T. A. (1997). Relationship between consci-
and construct validity. Journal of Applied Psychology, 84, 155-166. entiousness and learning in employee training: Mediating influences of
 BIG FIVE AND JOB PERFORMANCE 879

self-deception and self-efficacy. Journal of Applied Psychology, 82, mance in the European community. Journal of Applied Psychology, 82,
764 -773. 30-43.
Motowidlo, S. J., & Van Scotter, J. R. (1994). Evidence that task perfor- *Salgado, J. F., & Rumbo, A. (1997). Personality and job performance in
mance should be distinguished from contextual performance. Journal of financial services managers. International Journal of Selection and
Applied Psychology, 79, 475-480. Assessment, 5, 91-99.
Mount, M. K., & Barrick, M. R. (1995). The Big Five personality dimen- *Schmit, M. J., Motowidlo, S. J., Degroot, T. G., Cross, T. C., & Kiker,
sions: Implications for research and practice in human resources man- D. S. (1996, April). Explaining the relationship between personality and
agement. In K. M. Rowland & G. Fen-is (Eds.), Research in personnel job performance. In J. M. Collins (Chair), Personality predictors of job
and human resources management (Vol. 13, pp. 153-200). Greenwich, performance: Controversial issues. Symposium conducted at the annual
CT: JAI Press. convention for the Society of Industrial and Organizational Psychology,
*Mount, M. K., Barrick, M. R., & Stewart, G. L. (1998). Five-factor model San Diego, CA.
of personality and performance in jobs involving interpersonal interac- Schmitt, N. W., Gooding, R. Z., Noe, R. A., & Kirsch, M. (1984).
tions. Human Performance, 11, 145-165. Meta-analyses of validity studies published between 1964 and 1982 and
*Mount, M. K., Barrick, M. R., & Strauss, J. P. (1994). Validity of the investigation of study characteristics. Personnel Psychology, 37,
observer ratings of the Big Five personality factors. Journal of Applied 407-422.
Psychology, 79, 272-280. Schmitt, N., Rogers, W., Chan, D., Sheppard, L., & Jennings, D. (1997).
Murphy, K. R., & Shiarella, H. (1997). Implications of the multidimen- Adverse impact and predictive efficiency of various predictor combina-
sional nature of job performance for the validity of selection tests: tions. Journal of Applied Psychology, 82, 719-730.
Multivariate frameworks for studying test validity. Personnel Psychol-
Steiner, D. D., & Gilliland, S. W. (1996). Fairness reactions to personnel
ogy, 50, 823-854.
selection techniques in France and the United States. Journal of Applied
Ones, D. S., Mount, M. K., Barrick, M. R., & Hunter, J. E. (1994).
Psychology, 81, 134-141.
Personality and job performance: A critique of the Tett, Jackson, and
*Stewart, G. L. (1996). Reward structure as a moderator of the relationship
Rothstein (1991) meta-analysis. Personnel Psychology, 47, 147-156.
between extraversion and sales performance. Journal of Applied Psy-
*Piedmont, R. L., & Weinstein, H. P. (1994). Predicting supervisor ratings
chology, 81, 619-627.
of job performance using the NEO Personality Inventory. Journal of
*Stewart, G. L. (1997, August). Applicants versus incumbents: Assessing
Psychology, 128, 255-265.
the impact of validation design on personality research. In N. Schmitt
*Robie, C., & Ryan, A. M. (1998, April). Effects of non-linearity and
(Chair), Relating personality to job attitudes and job performance.
heteroscedasticity on the validity of conscientiousness in predicting
Symposium conducted at the 1997 Annual Meeting of the Academy of
overall job performance. Paper presented at the annual conference of the
Management, Boston.
Society for Industrial and Organizational Psychology, Dallas, TX.
*Rose, R. M., Fogg, L. F., Helmreich, R. L., & McFadden, T. J. (1994). *Stewart, G. L., & Carson, K. P. (1995). Personality dimensions and
Psychological predictors of astronaut effectiveness. Aviation, Space, and domains of service performance: A field investigation. Journal of Busi-
Environmental Medicine, 65, 910-915. ness and Psychology, 9, 365-378.
Rosse, J. G., Miller, H. E., & Barnes, L. K. (1991). Combining personality *Stewart, G. L., Carson, K. P., & Cardy, R. L. (1996). The joint effects of
and cognitive ability predictors for hiring service-oriented employees. conscientiousness and self-leadership training on employee self-directed
Journal of Business and Psychology, 5, 431-445. behavior in a service setting. Personnel Psychology, 49, 143-164.
Rosse, J. G., Miller, J. L., & Stecher, M. D. (1994). A field study of job Tett, R. P., Jackson, D. N., & Rothstein, M. (1991). Personality measures
applicants' reactions to personality and cognitive ability testing. Journal as predictors of job performance: A meta-analytic review. Personnel
of Applied Psychology, 79, 987-992. Psychology, 44, 703-742.
Rosse, J. G., Stecher, M. D., Miller, J. L., & Levin, R. A. (1998). The Tett, R. P., Jackson, D. N., Rothstein, M., & Reddon, J. R. (1994).
impact of response distortion on preemployment personality testing and Meta-analysis of personality-job performance relations: A reply to
hiring decisions. Journal of Applied Psychology, 83, 634-644. Ones, Mount, Barrick, and Hunter (1994). Personnel Psychology, 47,
Rothstein, H. R. (1990). Interrater reliability of job performance ratings: 157-172.
Growth to asymptote level with increasing opportunity to observe. *Van Scotter, J. R., & Motowidlo, S. J. (1996). Interpersonal facilitation
Journal of Applied Psychology, 75, 322-327. and job dedication as separate facets of contextual performance. Journal
Ryan, A. M., Ployhart, R. E., & Friedel, L. A. (1998). Using personality of Applied Psychology, 81, 525-531.
testing to reduce adverse impact: A cautionary note. Journal of Applied Whitener, E. M. (1990). Confusion of confidence intervals and credibility
Psychology, 83, 298-307. intervals in meta-analysis. Journal of Applied Psychology, 75, 315-321.
*Rybicki, S. L., & Klippel, D. C. (1997, April). Exploring the impact of
personality syndromes on job performance. Poster session presented at
the 12th annual convention for the Society of Industrial and Organiza- Received N o v e m b e r 30, 1998
tional Psychology, St. Louis, MO. Revision received N o v e m b e r 8, 1999
Salgado, J. F. (1997). The five factor model of personality and job perfor- Accepted N o v e m b e r 11, 1999 •

View publication stats